Apart from other derived anatomical features, such as the presence of eleven instead of ten primaries, an outer calcium phosphate layer of the eggshell, and a unique taxon of Cestodes, flamingos and grebes share a number of derived osteological characters. In addition to the characters optimized as synapomorphies of grebes and flamingos in the analysis of Mayr (2004) and those subsequently identified by Manegold (2006), there are previously unrecognized derived characters, which bear on the affinities of Juncitarsus. Four synapomorphies support a clade including Juncitarsus, Phoenicopteriformes, and Podicipediformes: •

cervical vertebrae with well-developed processus ventrolaterales (Fig. 3; Manegold, 2006). Livezey (2011: 394) assumed that the “process in flamingos illustrated by Manegold (2006) is not homologous with the facies articularis of ‘processus postlateralis’ of grebes, which is associated with a unique insertio accessorius [sic] tendinis m. longus colli ventralis (Zusi and Storer, 1969)”. It is, however, unknown, whether the latter authors did indeed study Phoenicopteriformes with regard to this character, and occurrence of a derived myological feature does not necessarily preclude homology of the osteological structures. In Juncitarsus, processus ventrolaterales are present (Fig. 3A);

The following seven characters are synapomorphies of Mirandornithes to the exclusion of Juncitarsus: •

at least four thoracic vertebrae fused and forming a notarium (Mayr, 2004). Juncitarsus lacks extensively fused thoracic vertebrae (Olson and Feduccia, 1980 and Peters, 1987), and only two vertebrae form a notarium in J. merkeli (Fig. 6);

The status of two derived characters shared by Phoenicopteriformes and Podicipediformes cannot be evaluated for Juncitarsus, and these either constitute synapomorphies of Juncitarsus and Mirandornithes, or apomorphies of Mirandornithes: •

unusually high number of at least 23 praesacral vertebrae (Mayr, 2004). Livezey (2011: 394) criticized this character because “the states were delimited by unequally wide ranges of counts, and ordering inflated the influence of the character”. These possible analytical issues do, however, not challenge the status of this feature as a synapomorphy of grebes and flamingos, because the high vertebral count found in these taxa clearly is derived within Neoaves and occurs in very few other groups (Mayr and Clarke, 2003). The number of praesacral vertebrae of Juncitarsus is unknown, as an exact count is not possible in the J. merkeli holotype;

Apart from two plesiomorphic features, which distinguish Juncitarsus from charadriiform birds (humerus with single pneumatic foramen and without processus supracondylaris dorsalis), Olson and Feduccia (1980: 48) listed three derived characters in support of phoenicopteriform affinities of Juncitarsus: •

the fact that the cervical vertebrae are “modified similarly to modern Phoenicopteridae”;

None of these characters constitutes, however, strong evidence for a sister group relationship between Juncitarsus and Phoenicopteriformes. As detailed above, most of the derived features of the cervical vertebrae of flamingos are also found in grebes, and the marked scapulohumeralis cranialis scar is a synapomorphy of Mirandornithes. The eminentia intercotylaris of Palaelodus is not as strongly raised as in extant Phoenicopteridae (Fig. 5D–F) and more closely resembles that of Podicipediformes, so that the prominent eminentia of Juncitarsus and modern flamingos probably evolved convergently, possibly in functional correlation with the greatly elongated legs of these birds.

In summary, the osteology of Juncitarsus is in better agreement with a sister group relationship to Mirandornithes than with an assignment to Phoenicopteriformes (Fig. 8). Juncitarsinae Peters, 1987, which were considered to be a taxon within Phoenicopteriformes, is thus elevated to a family-rank taxon, Juncitarsidae (Peters, 1987).

Molecular data do not provide unequivocal evidence for the closest extant relative of the flamingo/grebe clade. In studies with a comprehensive taxon sampling that include β-fibrinogen sequences, Phoenicopteriformes and Podicipediformes are placed in a taxon termed “Metaves”, which was hypothesized to constitute one of the two major lineages of Neoaves (all extant birds except Palaeognathae and Galloanseres; Ericson et al., 2006, Fain and Houde, 2004 and Hackett et al., 2008). A metavian clade is, however, not obtained in analyses excluding β-fibrinogen sequences and, because of the lack of independent complementary evidence, likely to be an artifact of this gene (Mayr, 2011a, McCormack et al., 2013 and Morgan-Richards et al., 2008). Mayr (2011a) hypothesized close affinities between flamingos and grebes and the Madagascan Mesitornithidae (mesites), because combined sequence data from c-myc, RAG-1, myoglobin, and ornithine decarboxylase support a clade (Mesitornithidae + [Phoenicopteriformes + Podicipediformes]), whereas β-fibrinogen sequences result in sister group relationship between Mesitornithidae and Phoenicopteriformes (Ericson et al., 2006: Figs. ESM-4 and ESM-6; see also Kimball et al., 2013).

In analyses of nuclear sequences exclusive of β-fibrinogen data, Phoenicopteriformes and Podicipediformes group with either Otididae (bustards), Musophagidae (turacos), and Columbiformes (doves and sandgrouse; McCormack et al., 2013: Fig. 2A; see also Kimball et al., 2013: Fig. 5B), Eurypygidae (sunbittern) and Phaethontidae (tropicbirds; McCormack et al., 2013: Fig. 2B), or the clade (Mesitornithidae + Mirandornithes) results in a polytomy with Charadriiformes and core-Gruiformes (Rallidae, Heliornithidae, Psophiidae, Aramidae, and Gruidae) (Ericson et al., 2006: Fig. ESM-6). Analyses of nuclear gap characters identifies Pelecanidae (pelicans) as sister taxon of flamingos and grebes (Yuri et al., 2013: Fig. 3), whereas some mitochondrial data support sister group relationship between the Phoenicopteriformes/Podicipediformes clade and Charadriiformes (Brown et al., 2008: Fig. 2; Morgan-Richards et al., 2008 and Pratt et al., 2009: Fig. 1; see also Cracraft et al., 2004: Fig. 27.6). An analysis of nuclear PEPCK sequences, albeit with a restricted taxon sampling, also resulted in sister group relationship between Phoenicopteriformes and Charadriiformes (Sorenson et al., 2003: Fig. 3).

Obviously, thus, molecular data do not unambiguously identify the closest extant relative of Mirandornithes. However, although there is a fairly wide array of taxa that group with flamingos and grebes in sequence-based studies, Musophagidae, Columbiformes, Pelecanidae, and Phaethontidae are identified as close relatives only in singular data sets, and a sister group relationship is not supported by complementary evidence. That a close relationship between Charadriiformes and Mirandornithes is found in multiple analyses is particularly noteworthy, because charadriiform affinities of flamingos were already assumed before publication of the molecular results, with the shorebird-like morphology of Juncitarsus playing a central role in the establishment of this hypothesis (Olson and Feduccia, 1980 and Peters, 1987).

Peters (1987) noted that Juncitarsus and flamingos share with Charadriiformes (except Alcidae, see below) a derived morphology of the fourth toe, in which the fourth phalanx is shorter than the third (Fig. 7). This feature is present in only few other neornithine birds (Palaeognathae, Otididae, Sphenisciformes, Procellariiformes, Ardeidae, Balaenicipitidae, and Scopidae; Hesse, 1990: 13). In all other taxa, including Podicipediformes, the fourth phalanx is longer than the third, which is the plesiomorphic condition that also occurs in Archaeopteryx (Hesse, 1990). Based on the here proposed phylogenetic interrelationships of Juncitarsus, Phoenicopteriformes, and Podicipediformes (Fig. 8), it is equally parsimonious to assume that shortening of this phalanx occurred independently in Juncitarsus and Phoenicopteriformes, or that a short fourth phalanx is plesiomorphic for Mirandornithes and was secondarily elongated in Podicipediformes. That such can indeed occur is exemplified by Alcidae, which are phylogenetically nested within other Charadriiformes (e.g., Fain and Houde, 2007 and Mayr, 2011b), and which are the only charadriiform birds with a long fourth phalanx.

Peters (1987) further assumed that Juncitarsus had schizorhinal nostrils, which are indicative of rhynchokinesis, and which are a derived characteristic of some charadriiform and “gruiform” birds (e.g., Zusi, 1984). However, although the nostrils of Juncitarsus are very long as in, e.g., some species of Rallidae (Fig. 9), their caudal ends are not slit-like and do not reach the nasofrontal hinge as in truly schizorhinal birds. In any case, schizorhinal nostrils would not constitute unambiguous evidence for charadriiform affinities, because Charadriiformes include schizorhinal and holorhinal taxa, and from the phylogenetic interrelationships of the extant taxa it cannot be deduced, which condition was present in the stem species of the clade (Mayr, 2011b).

For grebes, “gruiform” affinities were also previously discussed (see Olson, 1985), and Zusi and Storer (1969) found that Podicipediformes share a derived morphology of the cervical musculus longus colli ventralis with the sister taxa (e.g., Ericson et al., 2006 and Hackett et al., 2008) Eurypygidae and Rhynochetidae (kagu). Manegold (2006) correlated this feature with a derived morphology of the vertebrae also present in flamingos, i.e., the presence of marked processus ventrolaterales; these likewise occur in Juncitarsus (Fig. 3A) and thus belong to the groundplan of the clade (Juncitarsus + Mirandornithes). Like mesites, the sunbittern and the kagu further share a notarium and derived properties of the β-fibrinogen gene with grebes and flamingos. An extensively fused notarium is, however, absent in Juncitarsus (Fig. 6), and thus probably evolved convergently in the clade (Eurypygidae + Rhynochetidae) and in Mirandornithes.

In overall morphology, e.g., with regard to the shape of the major wing bones and the short and plantarly deflected trochlea metatarsi II, Juncitarsus clearly corresponds better with Charadriiformes and core-Gruiformes than with Mesitornithidae, Eurypygidae, and Rhynochetidae. Juncitarsus and Mirandornithes further share with Charadriiformes and core-Gruiformes several derived characters that are absent in Mesitornithidae, Eurypygidae, and Rhynochetidae, such as a distinctly notched distal rim of the condylus medialis of the tibiotarsus (Fig. 4). One of the key apomorphies of grebes, the presence of lobed feet, is otherwise also only known from the “gruiform” sungrebes (Heliornithidae) and coots (Rallidae), and the charadriiform phalaropes (Phalaropidae), and may constitute a case of parallelism in closely related taxa.

Although the closest extant relatives of Mirandornithes can ultimately be identified only through congruent results of future analyses of molecular or morphological data, Charadriiformes remain among the candidate taxa, and it is to be hoped that future studies will yield data for a robust phylogenetic placement of Mirandornithes, Charadriiformes, and the putatively closely related “gruiform” birds.

Because of their highly divergent life habits and morphology, flamingos and grebes are certainly among the most interesting examples for the evolution of disparate morphologies among birds. It has been detailed previously that the very divergent morphologies of extant Phoenicopteriformes and Podicipediformes are bridged by the Palaelodidae, which combine flamingo-like skull features with a foot morphology reminiscent of that of grebes (Mayr, 2004). Palaelodids are known from very abundant material and especially the species Palaelodus ambiguus is represented by thousands of bones representing virtually every skeletal element (e.g., Cheneval and Escuillié, 1992, Milne-Edwards, 1871 and Worthy et al., 2010). Although the osteology of Palaelodus is comparatively well known, many features, especially of the skull, have not yet been evaluated in light of the novel hypothesis of a sister group relationship between flamingos and grebes. Such a revision of palaelodid morphology is beyond the scope of this study, but it should be noted that contra Mayr (2009), who hinted at the possibility that palaelodids are the sister taxon of Podicipediformes, phoenicopteriform affinities of Palaelodus are well supported by several shared derived features, including a deep lower mandible, the presence of occipital fontanelles on the cranium (Cheneval and Escuillié, 1992), the elongation of the cervical vertebrae, and the presence of a tubercle on the distal tibiotarsus, laterodistal of the pons supratendineus. Future studies will also have to evaluate the exact affinities of other taxa assigned to Palaelodidae, such as the New World Megapaloelodus, at least some species of which show plesiomorphic features suggesting a position outside Mirandornithes (i.e., a low number of vertebrae fused into the notarium and a truncate proximal plantar articulation facet of the trochlea metatarsi III; e.g., Agnolin, 2009).

Palaelodus exhibits swimming adaptations, such as a mediolaterally compressed tarsometatarsus and proximally elongated cristae cnemiales of the tibiotarsus (Mayr, 2004), which are absent in Juncitarsus. A number of other derived characteristics of Phoenicopteriformes and Podicipediformes, which are also absent in Juncitarsus and thus likely evolved after the split of the latter from the lineage leading to modern flamingos and grebes, may also be correlated with the transition from a wading to a swimming habit, such as the extensively fused notarium, the large patella, closure of the hypotarsal canals, and flattening of the ungual phalanges. The life habits of Juncitarsus are poorly known, but the extremely elongated legs suggest that it was a wading bird, which probed for food in a lacustrine environment (Peters, 1987). That Juncitarsus was not yet adapted for filter feeding is not only indicated by the shape of its long and pointed beak, but also by the presence of numerous gastroliths in the J. merkeli holotype (Fig. 4A), which suggest a diet consisting of harder items (Peters, 1987). Whereas the stem species of Mirandornithes was likely to be a bird with swimming adaptations, such were probably absent in the stem species of the clade (Juncitarsus + Mirandornithes) and evolved after the divergence of Juncitarsus and Mirandornithes (Fig. 8).

Removal of Juncitarsus from the stem lineage of Phoenicopteriformes is of significance for the calibration of molecular data, and currently there exists no unambiguous Eocene record of Phoenicopteriformes, whose earliest fossils date from the Early Oligocene (Mayr, 2009). The earliest described fossils of Podicipediformes are from the Early Miocene (Mayr, 2009). The fossil record would thus be broadly congruent with a divergence date of crown group Mirandornithes in the mid-Paleogene, some 45 million years ago, as suggested by a recent calibrated molecular phylogeny (Brown and van Tuinen, 2011).